Re: [petsc-users] Issue using multi-grid as a pre-conditioner with KSP for a Poisson problem

2017-07-10 Thread Jason Lefley
Thanks for the clarification. > On Jul 9, 2017, at 7:05 AM, Mark Adams wrote: > > > > What are the differences between gamg and mg with -pc_mg_galerkin option > (apart from the default smoother/coarse grid solver options, which I > identified by comparing the ksp_view

Re: [petsc-users] Understanding preallocation for MPI

2017-07-10 Thread Barry Smith
You might consider using http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/Mat/MatPreallocateInitialize.html and http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/Mat/MatPreallocateSetLocal.html#MatPreallocateSetLocal and friends These take out some of the busywork

Re: [petsc-users] Understanding preallocation for MPI

2017-07-10 Thread Stefano Zampini
Look at the function that begins at line 1050 as the link should redirect to Il 10 Lug 2017 11:18 AM, "Florian Lindner" ha scritto: Hey Stefano, Am 10.07.2017 um 16:36 schrieb Stefano Zampini: > Florian, > > Perhaps you might want to take a look at how this is done for

Re: [petsc-users] Understanding preallocation for MPI

2017-07-10 Thread Florian Lindner
Hey Stefano, Am 10.07.2017 um 16:36 schrieb Stefano Zampini: > Florian, > > Perhaps you might want to take a look at how this is done for MatIS > > https://bitbucket.org/petsc/petsc/src/f9d5775f43f69cbce5a7014a6ce3b24cc0e1214a/src/mat/impls/is/matis.c?at=master=file-view-default#matis.c-1050

Re: [petsc-users] Understanding preallocation for MPI

2017-07-10 Thread Stefano Zampini
Florian, Perhaps you might want to take a look at how this is done for MatIS https://bitbucket.org/petsc/petsc/src/f9d5775f43f69cbce5a7014a6ce3b24cc0e1214a/src/mat/impls/is/matis.c?at=master=file-view-default#matis.c-1050 Stefano Il 10 Lug 2017 10:23 AM, "Florian Lindner"

Re: [petsc-users] Understanding preallocation for MPI

2017-07-10 Thread Florian Lindner
Hey, one more question about preallocation: I can determine if a column index is diagonal or off-diagonal using that code if (col >= col_range_start and col < col_range_end) d_nnz[relative_row]++; else o_nnz[relative_row]++; My code, however uses index sets from which a